Developing Privacy-preserving AI Systems: The Lessons learned

Chen, Huili; Hussain, Siam U.; Boemer, Fabian; Stapf, Emmanuel; Sadeghi, Ahmad Reza; Koushanfar, Farinaz; Cammarota, Rosario

doi:10.1109/dac18072.2020.9218662

Cited by 4 publications

(4 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, a key disadvantage of this approach is the computational overhead. A recent survey [7] summarizes the various privacy-preserving primitives, such as multi-party computation, differential privacy, and federated learning, that could potentially be integrated into deep learning solutions, emphasizing that almost always a trade-off between security, performance and computational load should be considered.…”

Section: Introductionmentioning

confidence: 99%

Privacy-Preserving Medical Image Classification through Deep Learning and Matrix Decomposition

Popescu

Nita

Taca

et al. 2023

2023 31st Mediterranean Conference on Control and Automation (MED)

View full text Add to dashboard Cite

Deep learning (DL)-based solutions have been extensively researched in the medical domain in recent years, enhancing the efficacy of diagnosis, planning, and treatment. Since the usage of health-related data is strictly regulated, processing medical records outside the hospital environment for developing and using DL models demands robust data protection measures. At the same time, it can be challenging to guarantee that a DL solution delivers a minimum level of performance when being trained on secured data, without being specifically designed for the given task. Our approach uses singular value decomposition (SVD) and principal component analysis (PCA) to obfuscate the medical images before employing them in the DL analysis. The capability of DL algorithms to extract relevant information from secured data is assessed on a task of angiographic view classification based on obfuscated frames. The security level is probed by simulated artificial intelligence (AI)-based reconstruction attacks, considering two threat actors with different prior knowledge of the targeted data. The degree of privacy is quantitatively measured using similarity indices. Although a trade-off between privacy and accuracy should be considered, the proposed technique allows for training the angiographic view classifier exclusively on secured data with satisfactory performance and with no computational overhead, model adaptation, or hyperparameter tuning. While the obfuscated medical image content is well protected against human perception, the hypothetical reconstruction attack proved that it is also difficult to recover the complete information of the original frames.

show abstract

Section: Introductionmentioning

confidence: 99%

Privacy-Preserving Medical Image Classification through Deep Learning and Matrix Decomposition

Popescu

Nita

Taca

et al. 2023

2023 31st Mediterranean Conference on Control and Automation (MED)

View full text Add to dashboard Cite

show abstract

“…A recurrent yet motivating example is on medical research, where various hospitals would benefit from each others' patient records to learn how to detect and treat rare diseases [10]. However, current technology is not well prepared for the application of DL on shared data while keeping its privacy [11]. Therefore, it is essential to understand the current knowledge, the failures, and the future research lines in the field of Privacy-Preserving Deep Learning (PPDL).…”

mentioning

confidence: 99%

“…PPCT for DL have attracted the attention of cryptographers and computer scientists in recent years, with an increasing number of proposals published at a fast rate. However, despite the high amount of papers addressing this topic, the technology is still immature, and few actual deployments are being used in privacypreserving scenarios [11,13].…”

mentioning

confidence: 99%

See 1 more Smart Citation

SoK: Privacy-Preserving Computation Techniques for Deep Learning

Cabrero-Holgueras

Pastrana

2021

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

Deep Learning (DL) is a powerful solution for complex problems in many disciplines such as finance, medical research, or social sciences. Due to the high computational cost of DL algorithms, data scientists often rely upon Machine Learning as a Service (MLaaS) to outsource the computation onto third-party servers. However, outsourcing the computation raises privacy concerns when dealing with sensitive information, e.g., health or financial records. Also, privacy regulations like the European GDPR limit the collection, distribution, and use of such sensitive data. Recent advances in privacy-preserving computation techniques (i.e., Homomorphic Encryption and Secure Multiparty Computation) have enabled DL training and inference over protected data. However, these techniques are still immature and difficult to deploy in practical scenarios. In this work, we review the evolution of the adaptation of privacy-preserving computation techniques onto DL, to understand the gap between research proposals and practical applications. We highlight the relative advantages and disadvantages, considering aspects such as efficiency shortcomings, reproducibility issues due to the lack of standard tools and programming interfaces, or lack of integration with DL frameworks commonly used by the data science community.

show abstract

Enhancing User Privacy in Natural Language Processing (NLP) Systems

Behera,

Lakshmi,

Kondurkar

2023

Advanced Applications of Generative AI and Natural Language Processing Models

View full text Add to dashboard Cite

NLP has witnessed a remarkable improvement in applications, from voice assistants to sentiment analysis and language translations. However, in this process, a huge amount of personal data flows through the NLP system. Over time, a variety of techniques and frameworks have been developed to ensure that NLP systems do not ignore user privacy. This chapter highlights the significance of privacy-enhancing technologies (differential privacy, secure multi-party computation, homomorphic encryption, federated learning, secure data aggregation, tokenization and anonymization) in protecting user privacy within NLP systems. Differential privacy introduces noise to query responses or statistical results to protect individual user privacy. Homomorphic encryption allows computations on encrypted data to maintain privacy. Federated learning facilitates collaborative model training without sharing data. Tokenization and anonymization preserve anonymity by replacing personal information with non-identifiable data. This chapter explores these methodologies and techniques for user privacy in NLP systems.

show abstract

Developing Privacy-preserving AI Systems: The Lessons learned

Cited by 4 publications

References 27 publications

Privacy-Preserving Medical Image Classification through Deep Learning and Matrix Decomposition

Privacy-Preserving Medical Image Classification through Deep Learning and Matrix Decomposition

SoK: Privacy-Preserving Computation Techniques for Deep Learning

Enhancing User Privacy in Natural Language Processing (NLP) Systems

Contact Info

Product

Resources

About